THE ROLE OF TREE IMPROVEMENT PROGRAMS FOR EX SITU GENE CON- SERVATION OF COASTAL DOUGLAS-FIR IN THE PACIFIC NORTHWEST

Transcription

1 FOREST GENETICS 10(2): , 2003 THE ROLE OF TREE IMPROVEMENT PROGRAMS FOR EX SITU GENE CON- SERVATION OF COASTAL DOUGLAS-FIR IN THE PACIFIC NORTHWEST Sara R. Lipow ab, G. Randy Johnson c, J. Bradley St. Clair c & Keith J. Jayawickrama a a) Department of Forest Science, Oregon State University, Corvallis, OR 97331, U.S.A. C) USDA Forest Service, Forestry Sciences Laboratory, Corvallis, OR 97331, U.S.A. b) Present address: Oregon Department of Forestry, 2600 State Street, Salem, OR 97310, U.S.A. Received July 23, 2002; accepted July 7, 2003 ABSTRACT We enumerate the genetic resources for coastal Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco var. menziesii) in tree improvement programs in the Pacific Northwest USA and evaluate how they contribute to gene conservation of the species. The first-generation programs include over four million progeny from 33,928 selections planted on 999 test sites. Nearly 2,000 of the selections tested in the first-generation have been incorporated into second-generation programs. The first-generation tests serve as repositories for low-frequency alleles of potential future import and for variation in quantitative traits not presently under selection. Much of this genetic variation also resides in the many large in situ populations of Douglas-fir, but it can be more easily detected in genetic tests. Additionally, useful genetic variation can be more rapidly integrated into tree improvement programs with the least loss of genetic gain if it is identified in genetic tests rather than in in situ populations. Because first-generation genetic tests are of limited life, we develop a method for creating gene resource outplantings to retain for a longer period the valuable genetic resources in them. Key words: genetic tests, conservation genetics, Pseudotsuga menziesii, tree breeding INTRODUCTION Genetic diversity is important for sustainable forest management, and this has been recognized in international agreements (e.g., the Montreal Process, The size and breadth of a tree species gene pool directly affects its ability to evolve and thrive under changing environmental conditions. Without sufficient genetic diversity, species cannot adapt to new stresses such as disease and climate change, nor can tree breeders continue to achieve genetic improvement objectives. Gene conservation of forest trees can be accomplished through both in situ and ex situ methods. Most gene conservation plans stress in situ protection of natural populations in protected areas. Protected areas are cost effective since they simultaneously serve a variety of conservation objectives, including safeguarding the genetic resources of numerous species. Furthermore, populations in protected areas are subject to normal evolutionary processes. Ex situ methods, which include holding germplasm in cold storage or growing trees in off-site plantations, are emphasized for threatened populations or species that cannot easily be conserved in situ. Species undergoing domestication also have genetic resources conserved ex situ in the form of the hierarchy of populations that comprise the breeding and testing programs. Large, natural populations of coastal Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco var. menziesii) can be found in protected areas across the Pacific Northwest, resulting in strong in situ gene conservation (LIPOW et al. 2003; YANCHUK & LESTER 1996). In this paper, we explore the many ways that ex situ genetic resources also contribute to gene conservation in the region, with emphasis on western Oregon and Washington. There are several reasons why this is an interesting and relevant issue. First, Douglas-fir is the region s most important timber tree and a keystone species in many ecosystems, with the survival of many other plants and animals depending on it. Second, the sheer magnitude of the ex situ genetic resources, mainly open-pollinated first-generation progeny tests, is unparalleled for any commercial forest tree species. Third, the Pacific Northwest, more than any region in the United States, attracts scrutiny for evidence of responsible forest management, including management of its genetic resources. We therefore review the size and structure of the region s first- and second-generation breeding programs and show how they achieve specific gene conservation objectives. We also describe the advantages associated with maintaining genetic material A R B O R A P U B L I S H E R S 111

2 S. R. LIPOW ET AL.: TREE IMPROVEMENT PROGRAMS FOR EX SITU GENE CONSERVATION OF COASTAL DOUGLAS FIR previously tested in the breeding programs for the purpose of gene conservation. As genetic tests are of limited life, we outline a practical method that allows these valuable resources to be retained over the long term. EX SITU GENETIC RESOURCES OF COASTAL DOUGLAS-FIR Oregon and Washington: First-generation breeding programs Throughout western Oregon and Washington, forest land management organizations are engaged in coastal Douglas-fir tree improvement programs, most of which were founded on the progressive system first introduced by Roy Silen of the USDA Forest Service and Joe Wheat of the Industrial Forestry Association (SILEN 1966; SILEN & WHEAT 1979). In this system, firstgeneration parent trees were selected from wild stands, and a series of six to 12 open-pollinated half-sib progeny tests were used to generate information for roguing the corresponding first-generation seed orchards and for making second-generation selections. The parent trees came from small breeding zones that varied in size from about 12,000 to 80,000 ha, with an average elevation change of less than 305 m (QUAM, 1988). The breeding zones spanned most of the species regional range (Figures 1). They displayed little overlap since most organizations engaged in tree improvement participated in one or more of the 20 local Douglas-fir cooperatives comprising the Northwest Tree Improvement Cooperative (NWTIC). Seventy-two of the region s 124 separately managed first-generation programs belonged to the NWTIC (Table 1, Figure 1). The partitioning of selections into breeding zones effectively created multiple breeding populations (sensu NAMKOONG, 1976). The size and composition of most first-generation breeding populations were designed conservatively with the goal of maintaining genetic diversity. The breeding zones were kept small, so that selections would retain adaptations to local conditions and preserve coadapted gene complexes (SILEN & WHEAT 1979). Moreover, large numbers of parent trees were selected from each breeding zone to allow for continued selection without appreciably eroding genetic diversity in the breeding population (SILEN & WHEAT 1979). These factors led to the testing of about 31,178 selections in western Oregon and Washington alone, or an average of about 7.5 selected trees per 1000 ha (Table 1). The methods used to select parent trees for the breeding population also assured the capture of much of the species regional genetic diversity. Roughly 80 % of the selections, at least those in NWTIC programs, were not intensively selected plus trees. Rather, a roadside selection approach emphasized sampling trees from throughout the breeding zone that appeared healthy, vigorous, and well formed. The remaining NWTIC selections (approximately 20 %), as well as many selections in non-nwtic programs (e.g., Weyerhaeuser Company s), were identified using recommendations that considered diameter growth per zone space (STONECYPHER et al. 1996). Even with this refinement, the gain achieved was typically only a few percentage points, and significant bias in the genetic composition of the selections is therefore unlikely (STONECYPHER et al. 1996). In summary, the selection process, enormous number of parents trees, and stratification into many small breeding zones have assured an extremely broad genetic base both within and among breeding populations. The most extensive and significant of the ex situ genetic resources of coastal Douglas-fir in Oregon and Washington are the progeny tests of offspring from the ~31,178 parent trees. The tests are distributed on more than 861 test sites that sample a wide range of environmental conditions (Table 1). Because breeding zones are small, however, they remain in native environments and are subject to normal forces of natural selection. Number of offspring per test varies among programs. NWTIC tests typically include at least 12 trees per family grown at each of six to 12 test plantations; in other words, at least 72 offspring per parent tree. Many tests, especially older ones begun in the 1960s and 1970s, incorporate more individuals, and the total number of progeny in NWTIC programs surpasses a mind-boggling 2,700,000 trees, an average of roughly 130 progeny per parent tree (Table 1). Similar family sizes were used in non-nwtic programs, with the total number of progeny under test in western Oregon and Washington equaling about four million! Oregon and Washington: Advanced generation breeding programs In the past decade, cooperative second-generation breeding has gotten underway in most of western Oregon and Washington. Note, however, that despite testing of almost 11,000 first-generation parents in southwest Oregon (in the Oakridge, McKenzie, Roseburg, Medford, Powers and Gold Beach areas), there has been no move to proceed to cooperative secondgeneration breeding in this area. 112

3 FOREST GENETICS 10(2): , 2003 Figure 1. In western Oregon and Washington (USA), first-generation breeding zones for programs (a) in the Northwest Tree Improvement Cooperative and (b) of the USDA Forest Service, USDI Bureau of Land Management, Washington Department of Natural Resources, and industry (The Timber Company, Weyerhaeuser Company, and Simpson). The ongoing cooperative second-generation pro grams include only eight breeding zones, which encompass 72 first-generation programs (34 of which were NWTIC) (Figure 2). Each of the eight programs plans to test 51 to 379 second-generation families (1,947 total) created by crossing selections from within the breeding zone. These represent about 13 % of the firstgeneration families from the same area (Table 2). Weyerhaeuser Company, acting independently, is using only one low-elevation second-generation breeding zone for Washington, and it is incorporating selections from each of four corresponding first-generation breeding zones (WOODS 1993). The intense selection in the second-generation programs is expected to yield substantial genetic gains. Organizations restructured and enlarged breeding zones for the second generation because, although Douglas-fir populations possess unusually high levels of genetic variation and display considerable local adaptation (SILEN & MANDEL 1983; SORENSEN 1983; CAMPBELL ; CAMPBELL & SUGANO 1993), several lines of evidence suggested that the original breeding zones need not have been so small. Results from Weyerhaeuser Company s low elevation, firstgenerations tests, for example, showed the that same good families performed well across four first-generation breeding zones (STONECYPHER et al. 1996). Likewise, the performance of progeny in field tests in A R B O R A P U B L I S H E R S 113

4 S. R. LIPOW ET AL.: TREE IMPROVEMENT PROGRAMS FOR EX SITU GENE CONSERVATION OF COASTAL DOUGLAS FIR Table 1. Estimates of the numbers of parental selections included in first-generation tree improvement programs throughout western Oregon and Washington, British Columbia, and northern California, as well as the number of progeny tested. Most data are from a database of the NWTIC that underestimates slightly the real values. Program Number of Number of first- Number of Number of test breeding zones generation progenies plantations selections Oregon and Washington NWTIC USDA Forest Service a USDI Bureau of Land Management a 9 Oregon Department of Forestry a Washington DNR a Not available 21 Weyerhaeuser Company Not available Other industry California USDA Forest Service Not available 36 British Columbia Total > >999 a These organizations have independent programs in some regions, but in other regions they participate in NWTIC programs. Only Table the 2. Summary independent of programs the organization are listed of here. the ongoing second-generation NWTIC tree improvement programs for coastal Douglas-fir. Local families are tested within a breeding zone, whereas non-local families are tested across breeding zones Program Number of first- Number of first- Target-number of Target-number of generation programs generation second generation second generation local families non-local families Washington Coast Puget Sound Washington Cascades Vernonia Trask Coast Trask Inland South Central Coast North Oregon Cascades Total four NWTIC breeding zones and in a seed orchard correlated reasonably well (SILEN 1985). Additionally, families tended to rank similarly across sites within six breeding zones in Oregon; differences that did occur were generally not correlated with environmental differences among sites, suggesting that the firstgeneration breeding zones were not overly large (JOHNSON 1997). Despite evidence indicating that breeding zones could be expanded for the second-generation, few data address exactly how far families can be safely deployed. For this reason, NWTIC programs developed a breeding strategy emphasizing local crossing and wide testing. Second-generation crossing is concentrated within first-generation breeding zones, creating multiple populations within each second-generation breeding zone. Wide testing refers to the fact that each of these local crosses is planted on test sites throughout the new breeding zone. If future test data indicate that breeding zones were expanded too far, second-generation families can be subdivided into smaller breeding zones in the next generation. Because most crosses for the second-generation involved selections from the same first-generation breeding zone, subdividing could be done without losing any second-generation families because the parents involved in the cross were too far separated to be considered appropriate for any single 114

5 FOREST GENETICS 10(2): , 2003 Oregon and Washington: Seed stores In Oregon and Washington, extensive seed stores exist for coastal Douglas-fir. Databases obtained from the USDA Forest Service and Bureau of Land Management show that these agencies presently hold approximately 19,000 family seed lots in addition to bulked collections. This seed is stored for a variety of reasons, including for use in reforestation following logging or natural disturbance, research needs, and gene conservation. Most of the seed was collected from wild stands on federal lands, although some was obtained from seed orchards or trees on private lands. California Figure 2. In western Oregon and Washington (USA), second-generation breeding zones for programs in the Northwest Tree Improvement Cooperative. breeding zone. With subdividing, however, future programs might require infusion of additional selections from the first-generation programs. The multiple populations in the second-generation breeding zones also serve as sublines that should facilitate avoidance of inbreeding in seed orchards and in subsequent generations of breeding. In addition to the testing of local families over a wider breeding zone, elite crosses are being made across first-generation programs within a second-generation cooperative and the higher-ranked families from second-generation breeding programs are being tested in other second-generation cooperatives (non-local families) to provide the data necessary to determine if second-generation zones can be safely expanded even further. Tree improvement programs for coastal Douglas-fir in California also contain valuable genetic resources, although they are less extensive than programs in Oregon and Washington. For first-generation testing, the USDA Forest Service set up nine California breeding zones (KITZMILLER 1976), one of which was managed as a cooperative with industrial partners. These breeding zones are larger than the first-generation ones in Oregon and Washington. There are also several small industrial programs. The Forest Service selected 200 plus trees per breeding zone (1,800 total) and planted open-pollinated progeny from each at four test sites per breeding zone. The first progeny tests were initiated in 1976, and 12- to 20-year results are generally available. All parental selections have been archived in a clone bank that the Forest Service intends to maintain indefinitely to permit inclusion of all firstgeneration selections into second-generation programs. Second generation breeding has not begun, but, should it go forward, plans call for crossing the top 50 % of first-generation families in small partial diallels to test for general and specific combining ability and subjecting the remaining families to single pair matings followed by within-family selection (J. KITZMILLER, USDA Forest Service Regional Geneticist, pers. comm.). This strategy should maintain a broad genetic base in the second generation. British Columbia A coastal Douglas-fir tree improvement program run by the British Columbia Ministry of Forests comprises another substantial ex situ genetic resource. The program was designed with a single, large breeding zone for elevations below 700 m, because early provenance tests (CHING & HINZ 1978; ILLINGWORTH 1978) and a A R B O R A P U B L I S H E R S 115

6 S. R. LIPOW ET AL.: TREE IMPROVEMENT PROGRAMS FOR EX SITU GENE CONSERVATION OF COASTAL DOUGLAS FIR study of wide crossing (ORR-EWING et al. 1972) indicated relatively weak genetic differentiation among British Columbia populations. The first-generation program included plus trees identified in natural populations. Despite an intense selection process, low heritability of growth and form traits resulted in little genetic gain, and the selections are therefore expected to sample well genetic diversity in the breeding zone. A half-diallel design was used to progeny test 372 parent trees, with each family planted on 11 of the 88 test sites. Open-pollinated progeny from additional British Columbia trees and from 161 parent trees from western Washington were tested across five sites. The test results further affirmed that seed could be safely moved within the large breeding zone (WOODS 1993). The first-generation tests remain in place, and many of the original selections have been archived in clone banks (WOODS 1993). For the second generation breeding population, 450 parents were chosen based on forward selection of progeny from the half-diallels. The parents were divided into sublines of 12 to 16 clones to enable control of inbreeding in the associated seed orchards, although the sublining may be reevaluated as the tests generate data. Ex situ genetic resources for coastal Douglas-fir in other countries Coastal Douglas-fir has been widely exported to other parts of the world for use in plantation forestry, and genetic tests can be found in Europe, New Zealand, and Chile. An International Union of Forestry Research Organizations provenance trial set up from in 36 European countries, for example, tested 134 provenances of coastal Douglas-fir (KLEINSCHMIT & BASTIEN 1992). Many of these provenances still exist on at least one test site. Additional provenance trials and a breeding program that includes 1170 half-sib families from 35 provenances from coastal Oregon and Washington have been established in the United Kingdom, Belgium, France, Germany, Spain, and Italy, countries where Douglas-fir is widely grown commercially (J. C. BASTIEN, National Institute for Research (INRA), pers. comm.; J. KLEINSCHMIT, Lower Saxony Forest Research Institute, pers. comm.). In New Zealand, where Douglas-fir ranks as the second most important softwood, two provenance trials containing 82 sources were planted in 1957 and 1959 at 14 test sites, with at least 31 additional provenances from coastal Oregon and California included in more recently initiated tests (MILLER & KNOWLES 1994; C. B. LOW, New Zealand Forest Research Institute, pers. comm.). Many of the better provenances from these trials have been incorporated into a breeding program and archived in clone banks. In Chile, four recently established provenance tests include more than 700 Douglas-fir families from 25 provenances from the Pacific Northwest (P. RAMIREZ DE ARELLANO and C. BALOCCHI,Arauco,pers. comm.). Interestingly, in these non-native environments, individual families are being deployed over a much wider geographical and elevational range than in North America. TREE IMPROVEMENT PROGRAMS AND GENE CONSERVATION OBJECTIVES The ex situ genetic resources of coastal Douglas-fir in breeding populations and genetic tests in the Pacific Northwest are more extensive and complete than for any other forest tree species. The huge number of selections included go a long way towards meeting the three objectives most frequently associated with gene conservation (KANG 1979; KANG & NIENSTAEDT 1987): (1) to retain sufficient genetic variation in a breeding program to permit continued gain in subsequent generations; (2) to capture most rare (low-frequency) alleles; and (3) to preserve the normal adaptive potential of a population by maintaining variation in traits under quantitative genetic control (and not under strong directional selection). In forest trees, high levels of genetic gain can be realized for multiple generations in even relatively small breeding populations (e.g., initial sizes of ) (reviewed by WHITE 1992; JOHNSON et al. 2001). First-generation Douglas-fir test populations in the PNW average about 250 selections, with the smallest containing 70 selections; they should therefore permit continued genetic gain. Similarly, second-generation breeding populations average about 215 local families, with the smallest program (Puget Sound) including 90 local families. These population sizes should also be adequate unless new data indicate that some secondgeneration breeding zones require subdivision, in which case infusion of additional families from the first generations of testing may be necessary. Since relatively large population sizes are required to capture rare alleles, this gene conservation objective is usually accomplished by conserving trees in situ in protected areas (NATIONAL RESEARCH COUNCIL 1991). Yet, for Douglas-fir, the breeding and test populations are so extensive that they are expected to incorporate most alleles present in the natural populations from which they were drawn. While estimates vary, most suggest that a sample size of , which is smaller than that found in most Douglas-fir breeding programs, is sufficient to include, with 95 % probability, at least 116

7 FOREST GENETICS 10(2): , 2003 one copy of every neutral allele of frequency 0.05 or more (KANG 1979; NAMKOONG et al. 1988; FRANKEL et al. 1995). We should keep in mind, however, that for a rare allele to be useful for breeding for example because it confers disease resistance multiple copies are needed to avoid problems associated with monocultures and inbreeding (YANCHUK 2001). Furthermore, a single copy of a recessive allele could not be detected, as identification would require homozygous expression. Nevertheless, according to YAN CHUK S (2001) calculations, a breeding population of 250 selections (the average size of first-generation programs) is expected to include about twenty individuals expressing in either homozygous or heterozygous form 95 % of the dominant alleles of frequency 0.05 in the natural population. Fewer individuals expressing rare, recessive alleles would be found in the same breeding population, however, since only for allele frequencies greater than 0.30 can we expect 20 homozygous individuals. Very large sample sizes (approximately 10,000) are required to sample, with 95 % probability, twenty individuals with homozygous expression of a recessive allele of frequency In short, the very large first-generation testing populations are expected to include multiple individuals expressing all but the very rarest alleles, while the second-generation programs likely miss the expression of some rare, recessive alleles. A relatively large population size is also needed to capture and maintain over the long-term the normal genetic variance for quantitative traits present in a population that is not experiencing strong, directional selection. Again, estimates of the effective population size required vary. LYNCH (1996) proposed a target effective population of 1,000 based on an evaluation of how the forces of natural selection, random genetic drift, and mutation influence the genetic variance in a population. He showed that the genetic variance in populations of this size or larger is controlled mostly by selection-mutation balance, with genetic drift having a negligible effect. LANDE (1995), considering the population size at which genetic variation created by mutation is balanced with that lost by genetic drift, argued for an effective population size of These values apply only to populations subject to many generations of evolution. Thus, although Douglas-fir second generation breeding population sizes, which range from when non-local crosses are included, fall short of these targets, they nevertheless contain considerable genetic variation for quantitative traits. Moreover, since an average of 1,798 first-generation families were tested per second-generation breeding populations, we can be confident that these tests capture ex situ much of the quantitative trait variation present in natural populations. ADVANTAGES OF EX SITU GENE CONSERVA- TION IN TREE IMPROVEMENT PROGRAMS The hundreds of millions of coastal Douglas-fir trees growing in protected areas distributed throughout the Pacific Northwest provide for excellent in situ gene conservation (LIPOW et al. 2003; YANCHUK & LESTER 1996) and are of great value to the region s ecology and social systems. Nevertheless, ex situ genetic resources in tree improvement programs have the potential to offer some advantages with respect to utilizing valuable genetic variation. If, for example, a new disease were to strike Douglas-fir, or the impact of an existing disease increased (e.g., Swiss needle cast), screening existing genetic tests, rather than in in situ populations, could speed up the process of identifying genetic resistance and incorporating it into a breeding population. Once a resistance allele was detected in a tree in a genetic test, a breeder could turn to other family members for additional copies, since they would have high probability of also harboring the allele (e.g., for open-pollinated tests, an allele inherited from a parent tree is expected in half its progeny). This would be especially helpful if the resistance allele was at low frequency or recessive. Moreover, by discovering resistance in an already tested population, breeders could minimize the gain lost when developing new breeding and production populations. Genetic tests also permit evaluation of the effectiveness of resistance or other traits, since repeatability can be examined over several sites. Additionally, as mature trees, saplings, and seedlings can differ in traits expression, assessment of certain traits might be most successful when screening mature trees. The advantages of decreasing the time needed for breeding, and the gains lost, apply to traits under quantitative genetic control as well as to those controlled by low-frequency alleles. The prospect of global climate change creates another incentive for valuing ex situ genetic resources in genetic tests in addition to those protected in situ. Mounting evidence suggests that rapid global climate change will alter the distribution of many forest trees, including Douglas-fir (e.g., THOMPSON et al. 1998). Should this occur, the survival of any given population might be determined by whether it possesses the genetic variation necessary to respond to new selection pressures. Armed with information about adaptive tolerances from genetic tests, however, forest geneticists may be able to actively move germplasm and facilitate the establishment (or reestablishment) of well-adapted populations. Indeed, if left in place, genetic tests will continue to provide worthwhile and relevant data for decades. The provenance trials and breeding programs for coastal Douglas-fir in Europe, New Zealand, and Chile A R B O R A P U B L I S H E R S 117

8 S. R. LIPOW ET AL.: TREE IMPROVEMENT PROGRAMS FOR EX SITU GENE CONSERVATION OF COASTAL DOUGLAS FIR also serve to conserve genetic resources and may someday be beneficial to the forest industry or to ecosystem preservation in the Pacific Northwest. Given its worldwide use as a plantation species, the first encounter Douglas-fir has with a future pest or pathogen, and the first uncovering of tolerance or resistance, could be in a non-native environment, as could the detection of variants for a variety of valuable traits. It is therefore conceivable that improved Douglas-fir germplasm now in another country may someday be returned to North America for restoration or commercial purposes. Indeed this process is underway for noble fir (Abies procera Rehd.); stock native to Oregon and Washington and subjected to several cycles of breeding for bough traits in Denmark is being returned to the United States, where interest in the bough industry is expanding and breeding is less advanced (LIPOW et al. 2001). Genetic variation cannot be easily detected in seed stores, since they first require outplanting. For this reason, and because they are not subject to the normal evolutionary processes experienced by trees in the ground, the seed stores for Douglas-fir are less useful for gene conservation than trees in progeny tests or provenance trials. All the same, they contain considerable genetic variation that could be tapped into if needed. LONG-TERM CONSERVATION OF GENETIC RESOURCES REPRESENTED IN FIRST-GEN- ERATION GENETIC TESTS The gene conservation value of the ex situ genetic resources for coastal Douglas-fir hinges, in part, on maintenance of the first-generation progeny tests, since the second-generation programs alone are too small to accomplish the objectives of conserving most rare alleles and providing the full benefits associated with breeding for new traits or responding to climate change. The first-generation progeny tests, however, were not designed to last indefinitely. As they are often measured only through age 15, many have fulfilled or will soon fulfill their primary objectives of contributing data to rank the performance of parental selections. Left alone, self-thinning will naturally occur in the maturing tests typically starting at age years, depending on site conditions, elevation, and plantation spacing resulting in a decline in stand health, loss of test balance, and possibly loss of some slower growing families. In any case, measures are in place in all regions engaged in Douglas-fir tree improvement that aim to hold onto first-generation tests and selections, at least for a few decades. The NWTIC established guidelines designed to preserve enough first-generation progeny and test sites to allow for future assessment of new traits or reassessment of current traits and to maintain selections so that additional first-generation selections are available for inclusion into advanced generations should subdivision of second generation breeding zones be necessary (J. DANIELS,previousNWTICDirector, pers. comm.). The guidelines called for systematic thinning of three to four test sites per breeding zone to retain a more or less balanced test with about six trees per site for most families. Removing alternate diagonals is usually sufficient for this purpose. The remaining three to nine plantations per breeding zone are to be left standing, although they may be thinned with less concern about family balance. In all test sites, the goal is to hold onto all advanced generation selections. To date, these guidelines have been fairly well implemented (D. CRESS, Oregon State University, pers. comm.). Most tests were established between 1969 and 1993 and, because thinned tests are expected to remain viable until stands are 40 to 60 years old, we expect that the thinning guidelines will preserve % of the progeny, representing >95 % of the first-generation parents, until at least Importantly, successful maintenance of tests sites will require continued coordination, planning and monitoring so that stands are not inadvertently harvested or otherwise harmed. In California, first-generation progeny tests were established at tight spacing (6 6 ft) and usually required two thinnings before age 15. An interlocking block design allowed two out of three rows to be removed without loss of families or balance. Many tests now need additional thinning, which is expected to remove entirely about half of the most poorly performing FAMILIES (J. KITZMILLER, USDA Forest Service regional geneticist, pers. comm.). For use in a potential second-generation, the agency aims to preserve families per test site and trees per family, with at least half of the families common to all four test sites in the breeding zone. No family will be wholly lost, however, as all parental selections have been archived in a clone bank. Likewise, in British Columbia, most selections have been placed in clone banks. ESTABLISHING GENE RESOURCE OUTPLAN- TINGS For application when first-generation progeny tests mature to the point that they cannot be maintained, companies and agencies could consider setting up gene resource outplantings. These outplantings would perpetuate genes from previously tested families not 118

9 FOREST GENETICS 10(2): , 2003 included in the second-generation as an insurance policy against future changes in breeding objectives (such as selecting for resistance to a new disease). To minimize the costs associated with setting up outplantings and their size, an appropriate strategy might be to sample a fraction (say 10 or 20 %) of the next best performing families not included in the second-generation. The progeny in a first-generation test represent the F1 generation of the original parental selections (P). Open-pollinated seed could be collected from one F 1 individual per first-generation family and outplanted in rows, thereby creating an F 2 generation. Planting in rows would facilitate site establishment, make for easy demarcation of family identity, and keep costs down. Within an outplanting, replication of family rows would be unnecessary, because the primary objective is to preserve families, not to test them, although replication at multiple sites (two or three) is necessary to guard against catastrophes. A family size of 20 would ensure the inclusion of nearly all allelic variation in the F 1 selection, as it leads to a very high probability probably (0.998) of sampling both alleles at a heterozygous locus. Therefore, assuming no inbreeding, the F 2 family, like the F 1 individual, would include, identically by descent, about half of the alleles present in the parental selection (P). Even more genes from the parental selections would be represented in the gene resource outplantings because many F 1 individuals were likely pollinated by other F 1 individuals in the progeny test. With respect to choosing specific F 1 trees for seed collection, preference would be given to superior genotypes within a family. Because within-family heritabilities tend to be low (<0.20) in Douglas-fir (JOHNSON et al. 1997), any effects of within-family selection on allelic frequencies are likely to be small. Nevertheless, if growth were an adaptive trait, mild selection on growth would mimic natural selection and potentially increase slightly the frequency of important alleles needed in the future. Alternatively, the frequency of favorable alleles would be reduced if the favorable alleles were associated with slower growth or poorer form, but only by a small amount. Furthermore, the gain lost when establishing a new breeding population could be minimized somewhat by starting with an improved gene resource population produced by within-family selection. Gene resource outplantings extend some of the benefits associated with progeny tests for gene conservation. Screening for discernable differences among families would be possible and could be done easily in outplantings where families are planted in rows. While family differences are unlikely to be evident for many characteristics, the outplantings might provide important observations on major effects such as disease resistance or adaptation to a significant change in climate. They would also make possible within-family selection for future breeding or gene resource populations, and, because family rows allow for outcrossing, open-pollinated seed could be used following withinfamily selection without too much concern for inbreeding. Nevertheless, even though this method is designed to minimize costs, collecting and processing seed and establishing outplantings would be expensive. Therefore, long-term maintenance of first-generations tests remains the most cost efficient and valuable form of ex situ gene conservation of coastal Douglas-fir. ACKNOWLEDGEMENTS This work was supported by the Pacific Northwest Forest Tree Gene Conservation Group, whose members include Boise Cascade Corp., Olympic Resource Management, Inc., Oregon Department of Forestry, Oregon State University, Plum Creek Timber Company, USDA Forest Service (Region 6, Pacific Northwest Research Station, and State and Private), USDI Bureau of Land Management, Washington Department of Natural Resources, and Weyerhaeuser Company. Data was generously provided by D. Cress (NWTIC), J. Kitzmiller (California), and R. Stonecypher (Weyerhaeuser Company) and C. Dean (Weyerhaeuser Company). D. Cress, A. Yanchuk, J. Kitzmiller, and J. Thomas provided helpful comments on earlier versions of this manuscript. REFERENCES CAMPBELL,R.K. 1986: Mapped genetic variation of Douglasfir to guide seed transfer in southwest Oregon. Silvae Genet. 35: CAMPBELL,R.K. 1991: Soils, seed-zone maps, and physiography: Guidelines for seed transfer of Douglas-fir in southwestern Oregon. For. Sci. 37: CAMPBELL,R.K. &SUGANO,A.I. 1993: Genetic variation and seed zones of Douglas-fir in the Siskiyou National Forest. Res. Pap. PNW-RP-461. USDA For. Serv., Portland, OR. CHING, K.K. & HINZ, P.N. 1978: Provenance study of Douglas-fir in the Pacific Northwest Region. III. Field performance at age twenty years. Silvae Genet. 27: FRANKEL, O.H., BROWN, A.H.D. & BURDON, J.J. 1995: The Conservation of Plant Biodiversity. Cambridge University Press, Cambridge. ILLINGWORTH, K. 1978: Douglas-fir provenance trials in coastal British Columbia Results to six years after planting. In: Proceedings of the IUFRO Meeting of WP S.2-02, 05, 06, 13, 14. pp British Columbia Ministry of Forests, Vancouver, Canada. JOHNSON, G.R. 1997: Site-to-site genetic correlations and their implications on breeding zone size and optimum number of progeny test sites for coastal Douglas-fir. A R B O R A P U B L I S H E R S 119

10 S. R. LIPOW ET AL.: TREE IMPROVEMENT PROGRAMS FOR EX SITU GENE CONSERVATION OF COASTAL DOUGLAS FIR Silvae Genet. 46: JOHNSON, G.R., SNIEZKO, R.A. & MANDEL, N.L. 1997: Age trends in Douglas-fir genetic parameters and implications for optimum selection age. Silvae Genet. 46: JOHNSON, G.R., ST. CLAIR, J.B. & LIPOW, S.R. 2001: Genetic conservation in applied tree breeding programs. In: Proceedings, of the International Conference on Ex situ and In situ Conservation of Commercial Tropical Trees (eds. B.A. Thielges, S.D. Sastrapradja, A. Rimbawanto). pp Yogyakarta, Indonesia, June Faculty of Forestry, Gadjah Mada University, Yogykarta. KANG, H. 1979: Long-term tree breeding. In: Proceedings of the 15th Southern Forest Tree Improvement Conference. pp USDA For. Serv., Macon, GA. KANG, H. & NIENSTAEDT, H. 1987: Managing long-term tree breeding stock. Silvae Genet. 36: KITZMILLER, J.H. 1976: Tree Improvement Master Plan for the California Regions. USDA For. Serv., San Francisco, CA. KLEINSCHMIT, J. & BASTEIN, J.C. 1992: IUFRO s role in Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) tree improvement. Silvae Genet. 41: LANDE, R. 1995: Mutation and conservation. Conserv. Biol. 9: LIPOW, S.R., J.B. ST. CLAIR & G.R. JOHNSON. 2001: Ex situ gene conservation for conifers in the Pacific Northwest. USDA For. Serv. Gen. Tech. Rep. PNW-GTR-528. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, Oregon. 54 p. LIPOW, S.R., VANCE-BORLAND, K., ST. CLAIR, J.B., HENDER SON & MCCAIN, C. 2003: Gap analysis of conserved genetic resources in Pacific Northwest conifers. Conservation Biology (in press). LYNCH, M. 1996: A quantitative-genetic perspective on conservation issues. In: Conservation Genetics: Case Histories from Nature (eds. J.C. Avise & J..L. Hamrick). pp Chapman and Hall, New York. MILLER, J.T. & KNOWLES, F. B.: Introduced forest trees in New Zealand: Recognition, role and seed source, Part 14. Douglas-fir Pseudotsuga menziessii (Mirb.) Franco. FRI Bulletin No Rotorua: New Zealand Forest Research Institute Limited, New Zealand. NAMKOONG, G. 1976: A multiple-index selection strategy. Silvae Genet. 25: NAMKOONG, G., KANG, H.C. & BROUARD, J.S. 1988: Tree Breeding: Principles and Strategies. Springer-Verlag, New York. NATIONAL RESEARCH COUNCIL. 1991: Managing Global Genetic Resources: Forest Trees. National Academy Press, Washington DC. ORR-EWING, A.L., FRASER, A.R. & KARLSSON, I. 1972: Interracial crosses with Douglas fir: Early field results. British Columbia For. Serv. Res. Notes No. 55. QUAM, R. (ed./compil.) 1988: Progress Report for Local Cooperative Programs. Northwest Tree Improvement Cooperative, Corvallis, OR. SILEN, R.R. 1966: A simple, progressive tree improvement program for Douglas-fir. Res. Note PNW-45. USDA For. Serv., Portland, OR. SILEN R.R. 1985: More correlations of parent vs. offspring, Snow Peak and Umpqua Coast Cooperatives. Tree Improv. News 49: SILEN, R.R. & MANDEL, N.L. 1983: Clinal genetic growth variation within two Douglas-fir breeding zones. J. For. 81: SILEN, R.R. & WHEAT, J.G. 1979: Progressive tree improvement program in coastal Douglas-fir. J. For. 77: SORENSEN, F.C. 1983: Geographic variation in seedling Douglas-fir (Pseudotsuga menziesii) from the western Siskiyou Mountains of Oregon. Ecology 64: STONECYPHER, R.W., PIESCH, R.F., HELLAND, G.G., CHAP MAN, J.G. & RENO, H.J. 1996: Results from genetic tests of selected parents of Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco) in an applied tree improvement program. For. Sci. 42 (Monograph 32):1 35. THOMPSON, R.S., HESTLER, S.W., BARTLEN, P.J. AND ANDER SON, K.H. 1998: A strategy for assessing potential future changes in climate, hydrology, and vegetation in the western United Sates. USGS Circular 1153, US GPO, Washington, DC. WHITE, T. 1992: Advanced-generation breeding populations: Size and structure. In: Proceedings of IUFRO Conference S , Breeding Tropical Trees Resolving Tropical Forest Resource Concerns Through Tree Improvement, Gene Conservation, and Domestication of New Species. pp CAMCORE, Raleigh, NC. WOODS, J.H. 1993: Breeding programs and strategies for Douglas-fir in North America. In: Breeding Strategies of Important Tree Species in Canada (eds. Y.S. Port & G.W. Adams), For. Can. Publ. Report M-X-186E. YANCHUK, A.D. 2001: A quantitative framework for breeding and conservation of forest tree genetic resources in British Columbia. Can. J. For. Res. 31: YANCHUK, A.D. & LESTER, D.T. 1996: Setting priorities for conservation of the conifer genetic resources of British Columbia. For. Chron. 72: